notionary 0.1.13__py3-none-any.whl → 0.1.15__py3-none-any.whl

notionary/page/notion_to_markdown_converter.py

@@ -0,0 +1,261 @@
+from typing import Dict, Any, List, Optional
+
+from notionary.elements.block_element_registry import (
+    BlockElementRegistry,
+)
+from notionary.elements.block_element_registry_builder import (
+    BlockElementRegistryBuilder,
+)
+
+
+class NotionToMarkdownConverter:
+    """Converts Notion blocks to Markdown text with support for nested structures."""
+
+    def __init__(self, block_registry: Optional[BlockElementRegistry] = None):
+        """
+        Initialize the NotionToMarkdownConverter.
+
+        Args:
+            block_registry: Optional registry of Notion block elements
+        """
+        self._block_registry = (
+            block_registry or BlockElementRegistryBuilder().create_standard_registry()
+        )
+
+    def convert(self, blocks: List[Dict[str, Any]]) -> str:
+        """
+        Convert Notion blocks to Markdown text, handling nested structures.
+
+        Args:
+            blocks: List of Notion blocks
+
+        Returns:
+            Markdown text
+        """
+        if not blocks:
+            return ""
+
+        markdown_parts = []
+
+        for block in blocks:
+            block_markdown = self._convert_single_block_with_children(block)
+            if block_markdown:
+                markdown_parts.append(block_markdown)
+
+        return "\n\n".join(filter(None, markdown_parts))
+
+    def _convert_single_block_with_children(self, block: Dict[str, Any]) -> str:
+        """
+        Process a single block, including any children.
+
+        Args:
+            block: Notion block to process
+
+        Returns:
+            Markdown representation of the block and its children
+        """
+        if not block:
+            return ""
+
+        block_markdown = self._block_registry.notion_to_markdown(block)
+
+        if not self._has_children(block):
+            return block_markdown
+
+        children_markdown = self.convert(block["children"])
+        if not children_markdown:
+            return block_markdown
+
+        block_type = block.get("type", "")
+
+        if block_type == "toggle":
+            return self._format_toggle_with_children(block_markdown, children_markdown)
+
+        if block_type in ["numbered_list_item", "bulleted_list_item"]:
+            return self._format_list_item_with_children(
+                block_markdown, children_markdown
+            )
+
+        if block_type in ["column_list", "column"]:
+            return children_markdown
+
+        return self._format_standard_block_with_children(
+            block_markdown, children_markdown
+        )
+
+    def _has_children(self, block: Dict[str, Any]) -> bool:
+        """
+        Check if block has children that need processing.
+
+        Args:
+            block: Notion block to check
+
+        Returns:
+            True if block has children to process
+        """
+        return block.get("has_children", False) and "children" in block
+
+    def _format_toggle_with_children(
+        self, toggle_markdown: str, children_markdown: str
+    ) -> str:
+        """
+        Format toggle block with its children content.
+
+        Args:
+            toggle_markdown: Markdown for the toggle itself
+            children_markdown: Markdown for toggle's children
+
+        Returns:
+            Formatted markdown with indented children
+        """
+        indented_children = self._indent_text(children_markdown)
+        return f"{toggle_markdown}\n{indented_children}"
+
+    def _format_list_item_with_children(
+        self, item_markdown: str, children_markdown: str
+    ) -> str:
+        """
+        Format list item with its children content.
+
+        Args:
+            item_markdown: Markdown for the list item itself
+            children_markdown: Markdown for item's children
+
+        Returns:
+            Formatted markdown with indented children
+        """
+        indented_children = self._indent_text(children_markdown)
+        return f"{item_markdown}\n{indented_children}"
+
+    def _format_standard_block_with_children(
+        self, block_markdown: str, children_markdown: str
+    ) -> str:
+        """
+        Format standard block with its children content.
+
+        Args:
+            block_markdown: Markdown for the block itself
+            children_markdown: Markdown for block's children
+
+        Returns:
+            Formatted markdown with children after block
+        """
+        return f"{block_markdown}\n\n{children_markdown}"
+
+    def _indent_text(self, text: str, spaces: int = 4) -> str:
+        """
+        Indent each line of text with specified number of spaces.
+
+        Args:
+            text: Text to indent
+            spaces: Number of spaces to use for indentation
+
+        Returns:
+            Indented text
+        """
+        indent = " " * spaces
+        return "\n".join([f"{indent}{line}" for line in text.split("\n")])
+
+    def extract_toggle_content(self, blocks: List[Dict[str, Any]]) -> str:
+        """
+        Extract only the content of toggles from blocks.
+
+        Args:
+            blocks: List of Notion blocks
+
+        Returns:
+            Markdown text with toggle contents
+        """
+        if not blocks:
+            return ""
+
+        toggle_contents = []
+
+        for block in blocks:
+            self._extract_toggle_content_recursive(block, toggle_contents)
+
+        return "\n".join(toggle_contents)
+
+    def _extract_toggle_content_recursive(
+        self, block: Dict[str, Any], result: List[str]
+    ) -> None:
+        """
+        Recursively extract toggle content from a block and its children.
+
+        Args:
+            block: Block to process
+            result: List to collect toggle content
+        """
+        if self._is_toggle_with_children(block):
+            self._add_toggle_header_to_result(block, result)
+            self._add_toggle_children_to_result(block, result)
+
+        if self._has_children(block):
+            for child in block["children"]:
+                self._extract_toggle_content_recursive(child, result)
+
+    def _is_toggle_with_children(self, block: Dict[str, Any]) -> bool:
+        """
+        Check if block is a toggle with children.
+
+        Args:
+            block: Block to check
+
+        Returns:
+            True if block is a toggle with children
+        """
+        return block.get("type") == "toggle" and "children" in block
+
+    def _add_toggle_header_to_result(
+        self, block: Dict[str, Any], result: List[str]
+    ) -> None:
+        """
+        Add toggle header text to result list.
+
+        Args:
+            block: Toggle block
+            result: List to add header to
+        """
+        toggle_text = self._extract_text_from_rich_text(
+            block.get("toggle", {}).get("rich_text", [])
+        )
+
+        if toggle_text:
+            result.append(f"### {toggle_text}")
+
+    def _add_toggle_children_to_result(
+        self, block: Dict[str, Any], result: List[str]
+    ) -> None:
+        """
+        Add formatted toggle children to result list.
+
+        Args:
+            block: Toggle block with children
+            result: List to add children content to
+        """
+        for child in block.get("children", []):
+            child_type = child.get("type")
+            if not (child_type and child_type in child):
+                continue
+
+            child_text = self._extract_text_from_rich_text(
+                child.get(child_type, {}).get("rich_text", [])
+            )
+
+            if child_text:
+                result.append(f"- {child_text}")
+
+    def _extract_text_from_rich_text(self, rich_text: List[Dict[str, Any]]) -> str:
+        """
+        Extract plain text from Notion's rich text array.
+
+        Args:
+            rich_text: List of rich text objects
+
+        Returns:
+            Concatenated plain text
+        """
+        if not rich_text:
+            return ""
+
+        return "".join([rt.get("plain_text", "") for rt in rich_text])

{notionary-0.1.13.dist-info → notionary-0.1.15.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: notionary
-Version: 0.1.13
+Version: 0.1.15
 Summary: A toolkit to convert between Markdown and Notion blocks
 Home-page: https://github.com/mathisarends/notionary
 Author: Mathis Arends

notionary-0.1.15.dist-info/RECORD

@@ -0,0 +1,56 @@
+notionary/__init__.py,sha256=sVqdwzMQcc9jf6FKi1qflilsx8rnWzluVhWewVi5gyI,717
+notionary/notion_client.py,sha256=NPPK7zId6EaC-hQeFJ7geaiROGtZfmc8cVP2nQez5DU,6040
+notionary/database/database_discovery.py,sha256=qDGFhXG9s-_6CXdRg8tMiwX4dvX7jLjgAUFPSNlYtlI,4506
+notionary/database/database_info_service.py,sha256=Ig6gx8jUSPYORJvfgEV5kV6t72pZQsWU8HPMqd43B-o,1336
+notionary/database/notion_database.py,sha256=RY5MlXNE5DVNWLC_Derljsz87ZMHkE-05Vgm80kvLxg,7250
+notionary/database/notion_database_factory.py,sha256=Af57yaUHidD8TKJ8uyXOc2nnqHm7on6VGFdDRjxiq9o,6692
+notionary/database/models/page_result.py,sha256=Vmm5_oYpYAkIIJVoTd1ZZGloeC3cmFLMYP255mAmtaw,233
+notionary/elements/audio_element.py,sha256=XLARz5zlPPW_Qof6uhcYXFmyYzyS1fLdxdfsvh6GMOs,5589
+notionary/elements/block_element_registry.py,sha256=2mEbCEIPRY15qk6NYcq9vf7Bq_5vuu1xbiJatoLFf7w,8530
+notionary/elements/block_element_registry_builder.py,sha256=xfZWJaamYCyw-aIU8TXTcAncVJaLpN7ioJRLYQxTbCo,11529
+notionary/elements/bookmark_element.py,sha256=83ciz2THxjeq7ofn-Xz9sG1Ifzm_gkIDmOo0A-pNsSo,8577
+notionary/elements/callout_element.py,sha256=K8yk7nE3WP8nskJKbMunkRDFlhSCXrONmYS6hELN6QE,5932
+notionary/elements/code_block_element.py,sha256=wbW_PfH6QBUFVfeELESuLiI5-GmLm2YUFP4xwFHgNV4,5173
+notionary/elements/column_element.py,sha256=F_hnaBtQbnfRiKRVfpo9X5aNiw75SxeHfWOiMyMPcBw,10706
+notionary/elements/divider_element.py,sha256=RCN87xFizAjKd8c_beywNW5lfXFFtOZPT8tyXgAh2D0,2776
+notionary/elements/embed_element.py,sha256=LZjbSfwq0v8NGzwfUXpnGwvJ34IjYDwZzqyxVd2Iqt8,4768
+notionary/elements/heading_element.py,sha256=GsfEg5XtohmtO8PBP9giezIg6pRWQ_CdPXjh7jOiytw,2756
+notionary/elements/image_element.py,sha256=663H_FzE_bsovps3uCV12trNTmMAWBu5Ko1tSBNu2V4,4845
+notionary/elements/list_element.py,sha256=-f4mPRPesqFYYXfiiqGpnADeAY2ZAF1sTDtkLcejvLg,4846
+notionary/elements/mention_element.py,sha256=N1AcE8daMs6sDJGidQ8vvtHj30Do3K7x-4lW2aAYthU,8274
+notionary/elements/notion_block_element.py,sha256=lLRBDXhBeRaRzkbvdpYpr-U9nbkd62oVtqdSe-svT4c,1746
+notionary/elements/paragraph_element.py,sha256=ULSPcwy_JbnKdQkMy-xMs_KtYI8k5uxh6b4EGMNldTk,2734
+notionary/elements/qoute_element.py,sha256=3I2a7KboPF5QF6afu99HSIa62YUNfDJ6oaSDgyc9NjA,9041
+notionary/elements/table_element.py,sha256=jCq1ZFNCdxYEI6ER21OZ64TYGjW162gH-A6DUbldxY4,11222
+notionary/elements/text_inline_formatter.py,sha256=FE_Sq2cozpu5RVtMbnPq21gD06UjH3LMRYr3s16JKYo,10606
+notionary/elements/todo_lists.py,sha256=PrLZdp1-q1gx7mBsL6NUQ5U18HxmkUw8yqVh9qJpArE,4233
+notionary/elements/toggle_element.py,sha256=5cyWjkBgJcKdhhdrOn4bEM8Dpc5gXWhVbM4qZm9FNYM,7242
+notionary/elements/video_element.py,sha256=uYCwFakna7pzpCDjjtxRhUkSkPfzJTttdKUKYSluqyw,6031
+notionary/exceptions/database_exceptions.py,sha256=I-Tx6bYRLpi5pjGPtbT-Mqxvz3BFgYTiuZxknJeLxtI,2638
+notionary/exceptions/page_creation_exception.py,sha256=4v7IuZD6GsQLrqhDLriGjuG3ML638gAO53zDCrLePuU,281
+notionary/page/markdown_to_notion_converter.py,sha256=wTkH7o6367IWBtSqBrldpKx4rxHli176QfWtAenyysQ,15067
+notionary/page/notion_page.py,sha256=KIjVeiMJGWWxR6ty1uuNvMoQf2IoRmSUxwMdDIyOu40,17635
+notionary/page/notion_page_factory.py,sha256=UUEZ-cyEWL0OMVPrgjc4vJdcplEa1bO2yHCYooACYC8,8189
+notionary/page/notion_to_markdown_converter.py,sha256=WHrESgMZPqnp3kufi0YB7Cyy8U1rwhJ0d4HHdBRfRdU,8053
+notionary/page/content/notion_page_content_chunker.py,sha256=xRks74Dqec-De6-AVTxMPnXs-MSJBzSm1HfJfaHiKr8,3330
+notionary/page/content/page_content_manager.py,sha256=jXWscZyiFNLGMiuJ8Da9P26q_9Hit9_G1j0rVDlOc5M,6224
+notionary/page/metadata/metadata_editor.py,sha256=61uiw8oB25O8ePhytoJvZDetuof5sjPoM6aoHZGo4wc,4949
+notionary/page/metadata/notion_icon_manager.py,sha256=ixZrWsHGVpmF05Ncy9LCt8vZlKAQHYFZW-2yI5JZZDI,1426
+notionary/page/metadata/notion_page_cover_manager.py,sha256=qgQxQE-bx4oWjLFUQvpXD5GzO1Mx7w7htz1xC2BOqUg,1717
+notionary/page/properites/database_property_service.py,sha256=AJuBGahbb53VQa6IGGHxBMoOgCy6vFZg08uR_eDjNUs,11570
+notionary/page/properites/page_property_manager.py,sha256=Xl8Cwn8WVszqpFXT_NvASkmP5igpCTEgRVhG_F45424,6914
+notionary/page/properites/property_formatter.py,sha256=d_Nr5XQxgjB6VIS0u3ey14MOUKY416o_BvdXjbkUNAQ,3667
+notionary/page/properites/property_operation_result.py,sha256=PhxHJJxxG2BdDl7aswhWnMSmf9RQtoinKkRHDoqxwCs,3913
+notionary/page/properites/property_value_extractor.py,sha256=1BfyCYrFzfIUmNTozavrLTjG--6P6Dy2tkewf6rHHwQ,2353
+notionary/page/relations/notion_page_relation_manager.py,sha256=D7JZJLXjX2Jn3CIseJxoMK9qL9gp88t4NmL9Ihu06eg,12682
+notionary/page/relations/notion_page_title_resolver.py,sha256=jUYsEkfyDgdh77oh2awYEB5g1vQqLBq6xYSXL-4uPH8,1722
+notionary/page/relations/page_database_relation.py,sha256=F9aGXFjjL8ZLNbfTGeGm_QAyXhz2AEOw7GgDLdprEcE,2313
+notionary/page/relations/relation_operation_result.py,sha256=NDxBzGntOxc_89ti-HG8xDSqfY6PwyGHKHrrKbCzNjM,5010
+notionary/util/logging_mixin.py,sha256=fKsx9t90bwvL74ZX3dU-sXdC4TZCQyO6qU9I8txkw_U,1369
+notionary/util/page_id_utils.py,sha256=EYNMxgf-7ghzL5K8lKZBZfW7g5CsdY0Xuj4IYmU8RPk,1381
+notionary/util/singleton_decorator.py,sha256=GTNMfIlVNRUVMw_c88xqd12-DcqZJjmyidN54yqiNVw,472
+notionary-0.1.15.dist-info/licenses/LICENSE,sha256=zOm3cRT1qD49eg7vgw95MI79rpUAZa1kRBFwL2FkAr8,1120
+notionary-0.1.15.dist-info/METADATA,sha256=uXefq6BdsjQnMAf34az8oWfMb2_6L068evq_G8UGWlE,6154
+notionary-0.1.15.dist-info/WHEEL,sha256=CmyFI0kx5cdEMTLiONQRbGQwjIoR1aIYB7eCAQ4KPJ0,91
+notionary-0.1.15.dist-info/top_level.txt,sha256=fhONa6BMHQXqthx5PanWGbPL0b8rdFqhrJKVLf_adSs,10
+notionary-0.1.15.dist-info/RECORD,,

notionary/converters/__init__.py

@@ -1,50 +0,0 @@
-# Import converters
-from .markdown_to_notion_converter import MarkdownToNotionConverter
-from .notion_to_markdown_converter import NotionToMarkdownConverter
-
-# Import registry classes
-from .registry.block_element_registry import BlockElementRegistry
-from .registry.block_element_registry_builder import BlockElementRegistryBuilder
-
-# Import elements for type hints and direct use
-from .elements.paragraph_element import ParagraphElement
-from .elements.heading_element import HeadingElement
-from .elements.callout_element import CalloutElement
-from .elements.code_block_element import CodeBlockElement
-from .elements.divider_element import DividerElement
-from .elements.table_element import TableElement
-from .elements.todo_lists import TodoElement
-from .elements.list_element import BulletedListElement, NumberedListElement
-from .elements.qoute_element import QuoteElement
-from .elements.image_element import ImageElement
-from .elements.video_element import VideoElement
-from .elements.toggle_element import ToggleElement
-from .elements.bookmark_element import BookmarkElement
-from .elements.column_element import ColumnElement
-
-default_registry = BlockElementRegistryBuilder.create_standard_registry()
-
-# Define what to export
-__all__ = [
-    "BlockElementRegistry",
-    "BlockElementRegistryBuilder",
-    "MarkdownToNotionConverter",
-    "NotionToMarkdownConverter",
-    "default_registry",
-    # Element classes
-    "ParagraphElement",
-    "HeadingElement",
-    "CalloutElement",
-    "CodeBlockElement",
-    "DividerElement",
-    "TableElement",
-    "TodoElement",
-    "QuoteElement",
-    "BulletedListElement",
-    "NumberedListElement",
-    "ImageElement",
-    "VideoElement",
-    "ToggleElement",
-    "BookmarkElement",
-    "ColumnElement",
-]

notionary/converters/notion_to_markdown_converter.py

@@ -1,45 +0,0 @@
-from typing import Dict, Any, List, Optional
-
-from notionary.converters.registry.block_element_registry import (
-    BlockElementRegistry,
-)
-from notionary.converters.registry.block_element_registry_builder import (
-    BlockElementRegistryBuilder,
-)
-
-
-class NotionToMarkdownConverter:
-    """Converts Notion blocks to Markdown text."""
-
-    def __init__(self, block_registry: Optional[BlockElementRegistry] = None):
-        """
-        Initialize the MarkdownToNotionConverter.
-
-        Args:
-            block_registry: Optional registry of Notion block elements
-        """
-        self._block_registry = (
-            block_registry or BlockElementRegistryBuilder().create_standard_registry()
-        )
-
-    def convert(self, blocks: List[Dict[str, Any]]) -> str:
-        """
-        Convert Notion blocks to Markdown text.
-
-        Args:
-            blocks: List of Notion blocks
-
-        Returns:
-            Markdown text
-        """
-        if not blocks:
-            return ""
-
-        markdown_parts = []
-
-        for block in blocks:
-            markdown = self._block_registry.notion_to_markdown(block)
-            if markdown:
-                markdown_parts.append(markdown)
-
-        return "\n\n".join(markdown_parts)